Sparsity-assisted adaptive chirp mode decomposition and its application in rub-impact fault detection

Abstract

The rub-impact is a familiar failure and its timely detection is vital to maintain the security of whole system. The change of dynamic behavior caused by rub-impact fault can be reflected by the fast-oscillating instantaneous frequency (IF) of the vibration signal. The recently developed adaptive chirp mode decomposition (ACMD) has exhibited satisfactory capability in extracting fast-oscillating IF. However, the ACMD is easily affected by the noise, which greatly limits its application scopes. To overcome this issue, a novel method named sparsity-assisted adaptive chirp mode decomposition (S-ACMD) is developed, which incorporates sparsity-assisted IF update scheme that fully exploits the sparse prior of fast-oscillating IF caused by rub-impact fault. Meanwhile, the parameter selection method for getting optimal results is described. The efficacy of the proposed S-ACMD is confirmed by the vibration signals from dynamic simulation and rotor test rig, and the results show that it has strong anti-noise ability and can accurately extract rub-impact fault related fast-oscillating IF even in noisy cases.